A number of sharp characteristic luminescence lines has been observed for GaAs doped with 3d transition-metal impurities in the near-infrared region, the origin being attributed to the zero-phonon intracenter transitions between the energy levels of the metal ions split by the crystal field of the GaAs lattice. It is also known that these luminescence lines are very sensitive to the surrounding field of the transition-metal impurities. The luminescence of Cr-doped GaAs has been most extensively studied, the spectrum revealing a very sharp luminescence line at 0.839 eV. In this paper we review the results on the successful applications of this Cr-related luminescence line to characterization of in-depth profiles of arsenic vacancy in thermally annealed GaAs, local strain field in In-doped GaAs and interface stress at heterostructures grown on Cr-doped GaAs substrate.

This paper addresses the diffusion and gettering of Cu and Au to internal cavities in Si introduced by H-implantation. Rutherford backscattering and channeling and cross-sectional transmission electron microscopy are the main analysis methods used. During annealing at temperatures and times typical of low temperature device processing conditions, we observe a transient gettering regime in which implanted Au and Cu segregate to cavities leaving metal concentrations in the Si lattice well below the solubility level. Longer times and/or higher temperatures are required for equilibrium to be reached. These results may have important implications for developing optimum gettering strategies during thermal processing of device structures.

Gettering is widely used for fabricating integrated circuits using Si substrates, and has great potential for solar cell fabrications as well. Recently available solar cell efficiency studies have shown the benefits of the wafer backside Al, attributable to effects of gettering, a wafer backside field, and passivation of grain boundaries and dislocations. In this paper, we report experimental results which showed unambiguously that Czochralski Si wafer bulk minority carrier diffusion lengths can be significantly improved due to gettering of impurities by wafer backside Al, which also provided a protection from environmental contamination.

A MBE growth procedure of epitaxial silicon layers is demonstrated which includes a special designed buried strained compositionally graded Si1−xGex layer. Upon thermal relaxation closed dislocation loops are formed in this Si1−xGex layer without altering the structure of the Si top layer. This dislocated layer is shown to getter contaminants in the Si top layer reducing the concentration of deep levels in this layer to ≈1 × 1012 cm−3.

The lateral motion of iron impurities was observed and studied in ptype iron contaminated silicon. The lateral diffusion was induced by and then measured using Schottky diodes with a special interdigitated fingers design. Capture of the impurities was done by diffusing to laterally placed dislocation loops formed by a self aligned ion implantation. Lateral changes in Fe concentration were determined using capacitance-voltage and deep level transient spectroscopy.

Using the diffusion-segregation equation, modeling and simulations of gettering Au (a substi-tutional-interstitial species in Si) away from the Si bulk have been performed. Three external gettering schemes have been considered: wafer frontside P indiffusion gettering, wafer backside Al deposition gettering, and a combination of the two processes. Under the same processing conditions, it has been shown that P indiffusion gettering is faster than Al gettering, but P gettering has a lower gettering capacity and is less stable than Al gettering for longer gettering times. The combined P and Al gettering process is as fast as P gettering in reaching an optimum gettered state, and possesses the capacity and stability of the Al gettering process.

Gettering efficiencies and stabilities of internal gettering sites for metallic impurities in high and low carbon doped silicon have been compared with ramped and standard two-step pre-annealing conditions. This study was intended to compare two proposed techniques to shorten the long low temperature nucleation step in the standard Hi-Lo-Hi internal gettering site formation treatment. Specifically, we compare the affect of carbon and a ramped annealing sequence on oxygen precipitate formation and gettering effectiveness. Our results show both techniques accelerate oxygen precipitation, however, only the low carbon ramped materials produced efficient and stable gettering sites. The high carbon materials did not with either annealing treatment. This disparity in performance is due to a difference in the oxygen precipitate’s strain field. The precipitates in the low carbon material possessed a high strain field with strain-induced defects while in the high carbon material they were strain-free with no defects. These results indicate the strain stabilizes the gettered impurity such that the gettering rate is increased and stability is enhanced.

Gettering of interstitial iron in p/p+ epitaxial silicon wafers during an integrated circuit device simulation annealing was examined. The results from the deep level transient spectroscopy and optical microscope examination suggest no correlation between the interstitial iron concentration in the epitaxial layer and the bulk microdefect density. An electron microscopic analysis revealed that gettering of interstitial iron took place at an oxide polyhedral precipitate located at the center of a bulk stacking fault. By comparing the results of iron implanted to non-implanted samples, it is concluded that iron gettering occurs via a phase transformation of Si02 to an α FeSi2 iron disilicide. In this study, it is suggested that gettering of iron atoms is fundamentally different from that of other transition metals where the silicide formation occurs predominantly along the Frank partial dislocations of the bulk stacking fault.

Gettering represents a standard procedure to remove harmful impurities from device active regions of a silicon wafer. An affordable technique for the creation of a defect free (denuded) zone is the introduction of gettering sites on the wafer back side via mechanical damage (i.e. lapping). However, optimizing the extrinsic gettering process requires a technique to quantify the extent of damage introduced through a typical back-side damaging step. Here, we present results from a study using triple axis x-ray diffraction. We characterized the level of damage present in wafers subjected to different lapping conditions. By comparing reciprocal space maps from samples that had undergone different lapping steps, we concluded that all the as-lapped samples showed similar levels of damage in terms of both strain variations and lattice tilts. However, the integrated diffuse scattering intensities decreased from medium to soft t.-double-soft lapping, which confirmed that the double soft lapping introduced the least damage. Typically, the damage step introduced a symmetrical distribution of both lattice strain and lattice tilts (mosaicity) in the surface layer. KOH etching was employed to determine the depth of the damaged layer. In all cases the as-lapped samples had integrated intensities two orders of magnitude greater than those of the etched samples. Analysis of the residual damage present in the lapped samples after chemical etching confirmed that the more aggressive lapping conditions introduced damage of greater depth.

Iron gettering in p-type CZ silicon at 600C has been investigated using deep level transient spectroscopy(DLTS), and the surface photovoltage(SPV) method. A quantitative analysis of the gettering kinetics of iron using the polycrystalline silicon film is presented for the first time. Depth profiles of iron concentration indicated a sharp gradient in the Fe concentration near the polycrystalline silicon/substrate interface. Concurrent decrease in the minority carrier diffusion length was also observed in the same region. The majority of iron gettering from the bulk silicon was found to be associated with the enhancement of the internal gettering.

The presence of small oxygen precipitates/nuclei generated by prolonged heat treatment at 600C was found to prevent regeneration of FeB pairs at the room temperature. On the other hand, large precipitates formed at lOOOC do not influence the diffusion or the recombination of Fei with B to form FeB pairs.

Various silicon samples, both single and polycrystalline, were intentionally contaminated and externally gettered using phosphorus, aluminum and co- phosphorus/aluminum gettering. Gettering efficiencies were quantified via diffusion length improvements. Structural characterization was used to correlate defects with low gettering efficiencies. External gettering was found to be particularly effective at recovering diffusion length in large grain polycrystalline silicon and solar grade single crystal silicon despite Fe contamination and high defect densities. One of two explanations is possible, 1) the structural defects are initially undecorated and are completely gettered after Fe contamination, or 2) metal decoration on as-grown structural defects are structurally and/or chemically different from intentional Fe decoration.

Large grain polycrystalline silicon wafers have been subjected to post-thermal annealing after a POCl3 pre-diffusion or after a phosphorus doped silica-film deposition (1019p/cm3- 2.1021p/cm3). The different doping levels are obtained by a dilution of the P-doped SOG (2.1021 at/cm3) in a undoped SOG solution .For the first time we have achieved the maximum of the gettering efficiency after post-thermal annealing. The best combination of post thermal cycle parameters and doping level improves the minority carrier diffusion length of quite (300% to 400%) for POCl3 pre-diffused samples and (200% to 275%) for spin-on P-doped (P-SOG) polycrystalline silicon.

Transition-metal-hydrogen complexes have been introduced into bulk Si samples that contained Pt, Au, or Rh by the indiffusion of hydrogen at 1250°C from H2 gas. The structure and electrical properties of a PtH2 complex in Si have been studied by vibrational spectroscopy and electron paramagnetic resonance (EPR). The PtH2 complex has been found to introduce two levels in the Si bandgap. There is one paramagnetic charge state for which EPR provides detailed structural information and two nonparamagnetic charge states. The hydrogen vibrations of all three charge states of PtH2 have been assigned. In addition to the PtH2 complex, the hydrogen vibrations of several additional complexes in Si samples that contain hydrogen and Pt, Au, or Rh have been identified.

Theory has made great progress during recent years in calculating the fundamental properties of monatomic hydrogen in crystalline silicon. By applying the DLTS and DDLTS method we use the hydrogen-carbon complex which consists of an electronically inactive carbon atom on a substitutional lattice site and a hydrogen atom near the bond-center position to detect theoretically predicted properties of hydrogen in silicon. The results of two independent experiments show that there exists a coupling of the electronic and structural properties of monatomic hydrogen, as predicted by theory.

The diffusivity of deuterium (D) at 250°C was determined in silicon samples grown by different techniques. It is found that the diffusivity increases with the growth speed, increase in carbon content and a decrease in oxygen concentration of the substrate. These growth conditions correlate well with the concentration of vacancy-type defects in the as-grown state. Hence, we conclude that a vacancy mechanism is responsible for low-temperature hydrogen diffusion in silicon. The highest diffusivity for hydrogen, calculated from these data, was found to be 3 × 10−7 cm2/s.

Si wafers subject to short-time (4–12 min.), low-temperature atomic hydrogen cleaning in an electron cyclotron resonance (ESR) plasma system have been annealed subsequently in the temperature range 300–750 °C for 20 mins. While only a small broad peak is seen immediately after hydrogenation, several pronounced and distinct majority carrier trap levels show up in deep level transient spectroscopy (DLTS) measurements of subsequently fabricated Schottky diodes on samples annealed at 450 °C and above. The concentrations of these deep levels reach a maximum at anneal temperatures around 500 °C and drop substantially beyond 750 °C. This phenomenon appears to be unrelated to the presence of oxygen in Si and is of potential importance in silicon processing technology.

We report studies of passivation of the gold center in silicon by hydrogen and lithium using deep level transient spectroscopy (DLTS), capacitance voltage (CV) profiling and secondary ion mass spectroscopy (SIMS). Both lithium and hydrogen are able to remove the electrical activity of the gold center from the silicon band gap but the passivation mechanisms are different. In the case of lithium the passivation is most likely due to a Coulomb attraction between lithium donors Li+ and gold acceptors Au−. No complex formation is observed between Li+ and Au0. In contrast, hydrogen is able to passivate the gold center without the need of opposite charge states of the species involved. Two Au-H complexes are observed, one (G) electrically active, and another (PA) passive. Based on the annealing kinetics of these complexes we propose that the active complex is a Au-H pair and that the passive complex contains two H atoms (Au-H2).

We report the first observation of a hydrogen complexed EP (E’δ) center. Comparing EP/H and E’γ/H electron spin resonance spectra, we provide the first strong evidence that EP (E’δ) centers do not have a delocalized structure.

The presence of H in polycrystalline silicon gives rise to new and hitherto unexpected phenomena. In this paper two of the most recent observations are reviewed: (i) Hydrogen-induced metastable changes of the dark conductivity due to the formation and dissociation of an electrically active H complex and (ii) the generation of acceptor states during prolonged exposure of poly-Si to monatomic H at elevated temperatures. The observed type conversion is clearly due to the diffusion of excess H from the plasma since it does not occur during exposure to other species such as oxygen.

The dominant pathways for hydrogen diffusion in poly-Si TFT’s are identified by analyzing the hydrogenation effect on the various device geometries. It is observed that the gate poly-Si thickness and channel width didn’t affect the hydrogenation. As the channel length was decreased down to 3 µm, threshold voltage was reduced and field effect mobility was increased significantly with hydrogeantion time. In the thick gate oxide (2000 Å, 4000 Å) poly-Si TFT’s, the device characteristics have been improved rapidly with hydrogenation time. The tail state density of thin gate oxide TFT wasn’t change by hydrgenation while that of thick gate oxide TFT was significantly reduced. Our experimental results may support the model that hydrogen atoms diffuse into the bulk of the active channel layer through the gate oxide and passivate the grain boundary and intragranular defects limitedly by gate oxide area.